Agglomeration, Related Variety, and Vertical Integration

Abstract

Several recent studies have investigated the relationship between the geographic concentration of production and vertical integration, based on the hypothesis that the spatial agglomeration of firms in the same industry facilitates input procurement, thereby reducing the degree of vertical integration. This article contributes to this debate in two ways: first, we focus on interindustry vertical integration, and second, we consider the effects on vertical integration of unrelated and vertically related variety at the local level. The latter was measured using information from input-output tables and captured the opportunities for outsourcing within the local system. A data set of 24,663 Italian business groups in 2001 was used to estimate Tobit models to investigate the influence of vertically related variety and other agglomeration forces on the degree of vertical integration of groups. We found that vertical integration is influenced by industry specialization at the local level and that higher vertically related variety reduces the need for firms to integrate activities, since they have more opportunities to acquire intermediate goods and services within the local system. We analyze the manufacturing and different macroareas and show that this relationship is also influenced by technology and differences in the organization of economic activities at the local level.

Keywords:

• vertical integration
• agglomeration
• related variety
• business groups

Acknowledgments

We thank Andrés Rodríguez-Pose and three anonymous referees for their valuable and constructive comments and suggestions. We also thank Roberto Ganau and Yanlei Zhang for commenting on a previous version.

Notes

1 In Italy, a province is an administrative entity between a municipality and a region. According to a 2001 industry census, Italy consists of 103 provinces.

2 Specialization, concentration, and agglomeration at the local level refer to different concepts. However, given the high level of empirical correlation between them, they are often used synonymously in the literature.

3 We thank an anonymous referee for suggesting this term.

4 At the plant level, Diez-Vial and Alvarez-Suescun (2010) considered the presence of adjacent stages of production within the same plant, and Ono (2007) looked at whether manufacturing establishments outsource specific business services. The majority of studies used firm-level accounting data and industry-level census data to calculate the Adelman index or some modified version of it (Holmes 1999; Li and Lu 2009).

5 The Italian input-output table uses the classification of economic activities NaceRev.1.1.

6 Frenken et al. (2007) were interested in assessing the level of knowledge spillovers between industries and considered v_ij as an index of the similarities between industries based on their input structure. The index is intended to capture technological similarities among sectors.

7 Northwest includes Piemonte, Lombardia, Valle d’Aosta, Liguria; Northeast-Center includes Trentino Alto-Adige, Veneto, Friuli Venezia Giulia, Emilia-Romagna, Toscana, Umbria, Marche, Lazio; South includes Abruzzo, Molise, Campania, Puglia, Basilicata, Calabria, Sicilia, and Sardegna.

8 For reasons of space, we did not include the tables with the results, but these tables are available from us on request.